Doctoral Research

PhD
RESEARCH

Scalable hybrid reliability frameworks blending edge computing and cloud AI for real-time equipment health monitoring in the mining sector.

THE RESEARCH PROBLEM

Mining operations face a fundamental contradiction in reliability engineering: the assets that most need continuous monitoring — draglines, thickeners, ball mills, conveyors — are often located in environments with limited connectivity, harsh conditions, and massive data volumes that overwhelm traditional cloud-only architectures.

Existing reliability frameworks treat edge and cloud as separate choices. My research argues they must be integrated — with intelligent partitioning of processing between edge devices at the asset and cloud platforms for fleet-wide pattern recognition, all grounded in rigorous RCM2 methodology rather than pure data-driven approaches.

The goal is a scalable hybrid reliability framework that works in the real conditions of Australian mining — intermittent connectivity, diverse equipment types, regulatory requirements, and the practical constraints of maintenance teams in the field.

RESEARCH OBJECTIVES

01
Integrated Reliability Model

Develop an integrated reliability model that works across heterogeneous equipment types — from rotating machinery to process instrumentation — within a single unified framework.

02
Edge-Cloud Architecture

Design and validate a hybrid edge-cloud infrastructure that partitions processing intelligently — real-time anomaly detection at the edge, fleet-wide pattern learning in the cloud.

03
Digital Twin Validation

Validate digital twin deployment on edge and hybrid systems, establishing performance benchmarks for accuracy, latency, and resource consumption in mining environments.

04
RCM2 Integration

Integrate RCM2 methodology as the analytical backbone of the framework — ensuring AI-generated maintenance recommendations are grounded in engineering consequence logic, not just statistical pattern matching.

05
Multi-Equipment Generalisation

Develop and validate algorithms for multi-equipment model generalisation — so the framework learns from failures across the fleet and transfers that knowledge to individual assets.

06
Industry Validation

Conduct comparative validation with real-world industrial partners, comparing framework performance against existing maintenance approaches on production mining equipment.

RESEARCH MILESTONES

Complete
Literature Review – Reliability Modelling Approaches

Comprehensive review of existing reliability frameworks, edge computing architectures, digital twin methodologies, and RCM2 applications in mining. Gaps identified in hybrid edge-cloud integration.

Complete
Simulation Environment for Initial Model Testing

Established simulation environment for initial model testing using synthetic and public datasets. Baseline performance metrics established for comparison with proposed framework.

Complete
Preliminary Findings – International Conference

Published preliminary findings at international conference on digital twins. Framework architecture presented and validated by peer review. Feedback incorporated into refined research design.

In Progress
Multi-Equipment Generalisation Algorithm

Optimising the algorithm for transferring reliability models across heterogeneous equipment types. Current focus on thickener, conveyor, and rotating machinery datasets.

In Progress
Hybrid Edge-Cloud Infrastructure Prototype

Deploying prototype hybrid infrastructure on test environment. Edge nodes running lightweight anomaly detection; cloud layer performing fleet-wide pattern analysis and RCM2 decision logic.

Planned
Real-World Industrial Validation

Comparative validation with industrial partners on production mining equipment. Framework performance benchmarked against existing maintenance approaches across multiple asset types.

Planned
Thesis Submission

Final thesis documenting the complete hybrid reliability framework, validation results, and contribution to reliability engineering literature.

FRAMEWORK LAYERS

The hybrid framework operates across four integrated layers — from physical assets at the edge to fleet-wide intelligence in the cloud.

🏭
ASSET LAYER

Physical mining equipment with embedded sensors — vibration, temperature, pressure, current. Raw condition data generated continuously in the field.

EDGE LAYER

On-site edge devices running lightweight ML models for real-time anomaly detection. Low latency, operates on limited connectivity, generates alerts without cloud dependency.

☁️
CLOUD LAYER

Fleet-wide pattern recognition, digital twin synchronisation, RCM2 decision logic, and long-term reliability trend analysis. Learns from failure events across all assets.

🤖
INTELLIGENCE LAYER

AI-powered maintenance recommendations grounded in RCM2 methodology. Consequence-first logic ensures recommendations are defensible engineering decisions, not statistical outputs.

INTERESTED IN COLLABORATION?

Research partnerships, industrial validation opportunities, or discussions on hybrid reliability frameworks — get in touch.

Get In Touch Read the Blog